Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
9
, No
.
3
,
Ma
rch
201
8
,
pp.
539
~
542
IS
S
N:
25
02
-
4752
,
DOI: 10
.11
591/
ijeecs
.
v9.i
3
.
pp
539
-
542
539
Journ
al
h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Improve
ment of
Garbag
e Man
ag
em
ent for NA
ND
Mem
ory
System
A.
Thi
le
ep
an
, S.
Ra
m
achan
dran
AM
ET
Univer
sit
y
,
Chenn
ai
,
Ind
i
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
2
5
201
7
Re
vised
Jan
3
0
, 201
8
Accepte
d
Fe
b
19
, 201
8
Rec
en
t
da
y
s inc
r
ea
sing t
he
use
of
fla
sh m
emory
d
evi
c
e
in
embedd
ed
s
y
stems
.
Diver
se
qualitie
s
of
NA
ND
bla
ze
re
collections
from
har
d
ci
rc
les
inc
lude:
a
constra
in
ed
squa
re
era
d
icate
chec
k,
the
inc
onc
ei
v
a
bil
ity
of
se
t
up
r
efr
esh,
and
as
y
m
m
et
r
y
in
oper
ation
gra
nu
la
rity
.
Along
th
ese
li
n
es
diff
er
ent
rubbish
ac
cumula
ti
on
pr
oce
dure
s
for
th
e
NA
ND
strea
k
re
col
l
ec
t
ions
have
bee
n
proposed.
In
an
y
ca
se
,
exi
sting
rubbish
a
cc
um
ula
ti
on
proc
edur
es
obstruc
t
square
wea
r
le
v
el
in
g
sinc
e
they
uti
li
z
e
a
sim
il
ar
te
chni
qu
e
for
both
hot
and
icy
info
rm
at
ion.
In
thi
s
p
ape
r
,
we
propose
eff
e
ct
iv
e
junk
accu
m
ula
ti
on
and
pie
c
e
a
dm
ini
strat
ion
strategie
s
to
enh
ance
pi
ece
wea
r
l
eve
l
ing
and
tra
sh
gat
her
ing
spee
d
.
Above
all
e
lse
,
informat
ion
is
ar
ra
ng
ed
int
o
thre
e
sorts
conc
urring
to
alter
a
ti
on
re
cur
r
en
ce
-
hot
informati
on,
cool
infor
m
at
ion,
an
d
warm
dat
a and
d
isti
nctive
sorts o
f
informati
on
are put
awa
y
in
v
arious
pie
c
es
.
The
de
le
t
e
cost
is
figure
d
consi
der
ing
informat
i
on
sort,
and
afterward
jun
k
gat
her
ing
is
per
f
orm
ed
for
the
hi
nder
s
whos
e
eradic
a
te
costs
surp
ass
the
li
m
it
este
em.
Further
m
ore
,
unique
square
re
cor
ds
are
m
ade
in
RAM
by
informati
on
sort
,
and
the
squar
es
are
orc
h
estrate
d
in
the
r
equ
est
of
the
ir
era
di
ca
t
e cost.
Ke
yw
or
d
s
:
Flash m
e
m
or
y
NAN
D
RAM
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
A.
Thile
epa
n
,
AMET
Un
i
versi
ty
, Ch
en
nai
,
India
.
1.
INTROD
U
CTION
Streak
rec
ollec
ti
on
s
are
util
iz
ed
tod
ay
in
dif
fer
e
nt
instal
le
d
fr
am
ewo
r
ks
,
f
or
e
xam
ple,
ce
ll
ph
ones
,
PDAs,
a
nd
Digital
Cam
eras.
The
blaze
m
em
or
y
is
rising
as
a
ty
pical
stock
pili
ng
ga
dg
et
that
re
places
ha
r
d
ci
rcles
since
it
has
m
any
pref
eren
ces
s
uch
a
s
li
tt
le
siz
e,
high
stre
ngth
,
a
nd
low
power
uti
li
zat
ion
.
An
ad
aptiv
e
stripin
g
arc
hit
ect
ur
e
for
flas
h
m
e
m
or
y
storag
e
syst
e
m
s
of
em
bedded
syst
e
m
s
and
Locali
ty
-
awar
e
sect
or
translat
ions
f
or
N
A
ND
flash
m
e
m
or
y
-
base
d
sto
rag
e
syst
e
m
s
are
ex
plained
by
[
1]
,
[
2]
.
Be
that
as
it
m
ay
,
NAN
D
streak
reco
ll
ect
ions
ha
ve
a
co
up
le
of
c
onfinem
e
nts
becau
se
of
t
heir
physi
cal
qu
al
it
ie
s.
I
n
th
e
first
place,
be
fore
ne
w
inf
orm
ation
can
be
com
po
se
d
in
a
ra
nge
with
inf
or
m
at
ion
save
d
m
on
ey
on
it
,
the
reg
i
on
m
us
t
be
erad
ic
at
ed.
Des
pite
the
fact
that
perusin
g
an
d
com
po
sing
s
hould
be
poss
ible
in
pag
e
un
it
s
,
erad
ic
at
in
g
m
us
t
be
pe
rfor
m
ed
in
s
qu
a
re units.
Com
po
sin
g
m
ist
akes
ha
ppen
in
a
s
quare
that
ha
s
s
urpass
ed
t
he
per
m
it
te
d
delet
e
ta
ll
y.
Re
al
-
t
i
m
e
gar
ba
ge
c
ollec
ti
on
for
fl
ash
-
m
e
m
or
y
stora
ge
syst
em
s
is
discusse
d
by
[3
]
.
Along
these
li
nes,
al
l
reg
io
ns
of
the
bl
aze
m
e
m
or
y
ou
ght
to
be
un
i
form
l
y
utilized;
so
m
et
hin
g
el
se,
the
acce
ssible
stora
ge
r
oo
m
can
qu
ic
kly
be
le
ssene
d.
T
o
co
nque
r
these
po
i
nts
of
c
onfine
m
ent,
NAND
strea
k
reco
ll
ect
ions
com
po
se
ne
w
inf
or
m
at
ion
in
an
al
te
rn
at
e
spa
ce
and
refute
old
inf
orm
ati
on
wh
e
n
m
od
ify
i
ng
inf
or
m
at
ion
. E
ndurance
enha
ncem
ent o
f fla
sh
-
m
e
m
or
y st
orage
syst
em
s is
d
esc
ribe
d by [
4].
Likewise,
to
s
ecur
e
a
dequate
fr
ee
pieces,
wh
e
n
ther
e
ar
e
sq
ua
res
in
w
hich
in
valid
and
le
giti
m
a
te
inf
or
m
at
ion
ar
e
blen
de
d,
the
le
gitim
at
e
info
rm
at
ion
are
m
ov
ed
to
e
xhaust
pieces
a
nd
existi
ng
pie
ces
are
erased
a
nd
tra
ns
f
or
m
ed
int
o
fr
e
e
s
qu
a
res.
This
proc
ed
ure
is
cal
le
d
"r
ubbis
h
gathe
ring."
Be
ca
us
e
strea
k
reco
ll
ect
ions
s
et
asi
de
m
or
e
oppo
rtu
nity
to
e
rad
ic
at
e
tha
n
to
peruse
an
d
ke
ep
i
n
to
uc
h
w
it
h,
it
is
im
per
at
ive
to
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vol
.
9
,
No.
3
,
Ma
rc
h
201
8
:
539
–
542
540
lim
it
the
erad
ic
at
e
nu
m
ber
when
pe
rfor
m
ing
rub
bish
accum
ulati
on
.
P
rio
rity
-
base
d
ga
rb
a
ge
colle
ct
ion
f
or
data
stora
ge
syst
em
s
an
d
L
og
-
Bl
oc
k
Ma
na
gem
e
nt
Sche
m
e
fo
r
MLC
NAN
D
Flash
Me
m
or
y
Stor
a
ge
Syst
e
m
s
are
discusse
d
by
[5
]
,
[6
]
.
Mo
re
over
,
the
pieces
to
be
d
el
et
ed
ought
to
be
ch
os
e
n
w
hile
con
side
rin
g
squa
re
wea
r
le
veling
a
nd
th
e
cost
of
rep
li
c
at
ing
s
ubsta
ntial
pag
es
to
a
nother
piece.
E
xi
sti
ng
waste
gat
her
i
ng
strat
egi
es
an
d
piece
adm
inistr
at
ion
te
c
hn
i
ques
can'
t
ensu
re p
ie
ce
wea
r
le
ve
li
ng
,
sin
ce
the
y
don'
t
con
s
ide
r
inf
or
m
at
ion
s
or
t
as
far
a
s
al
te
rati
on
recurre
nce.
T
o
def
eat
these
issues,
this
pa
pe
r
proposes
a
pro
fici
ent
re
f
us
e
gat
her
in
g
an
d
piece
adm
inist
rati
on
by
ta
king
into
record
i
nfor
m
at
ion
sort.
A
n
e
ff
ic
ie
nt
desi
gn
of
se
rial
and
pa
rall
el
m
e
m
or
y
and
Sm
art
m
e
m
or
y
al
loys as str
uc
tural c
om
po
sit
es are
desc
ribe
d by [
7]
,
[8
]
.
1.
Prop
os
ed
NA
ND Flas
h me
mor
y
m
anag
e
ment
:
This p
a
pe
r
pr
opose
s a ru
bb
is
h
accum
ulati
on an
d sq
ua
re adm
inist
rati
on
techn
i
qu
e
that co
ns
ide
rs wea
r
le
veling
wh
il
e
fund
am
ental
l
y
util
iz
ing
the
MOD
A
Page
Allocat
ion
Me
thod.
T
he
pro
po
se
d
te
ch
nique
s
insti
gate
wea
r
l
eveli
ng o
f
al
l
s
qu
a
res
by
c
hoosi
ng
hi
nd
e
rs
f
or j
un
k
acc
um
ulati
on
i
n
view
of v
ari
ou
s
crit
er
ia
by
inf
or
m
at
ion
sort.
Be
sides,
a
f
ree
square
assi
gn
m
ent
te
chn
i
qu
e
that
co
ns
i
der
s
wear
le
ve
li
ng
by
ind
e
pe
n
de
ntly
m
aking
fr
ee
s
quare
r
ec
ords b
y i
nfor
m
at
ion
s
or
t i
s e
xhibit
ed
.
a)
Genera
te
a fr
ee block
a
ll
oc
at
i
on
me
thod
Hinders
t
hat
st
or
e
hot
inf
orm
at
ion
,
frosty
in
form
ation
,
a
nd
war
m
inform
at
ion
(
w
hich
don'
t
ha
ve
a
place
with
hot
or
ic
y
inf
or
m
ation)
ha
ve
disti
nctive
pro
ba
bili
ti
es
of
bein
g
e
rad
ic
at
ed
.
I
n
this
m
ann
er,
di
ver
s
e
crit
eria
for
ap
portio
ning
f
ree
sq
ua
res
a
re
re
qu
i
red
w
hen
c
om
po
sin
g
this
inf
or
m
at
ion
.
T
he
pr
opos
e
d
strat
egy
or
c
hestrates
f
r
ee
sq
ua
res
in
the
S
m
ash
in
the
risin
g
r
equ
e
st
of
thei
r
erad
ic
at
e
ta
ll
y,
an
d
m
akes
three
arr
a
ng
em
ents
of
f
ree p
ie
ces, i.e.,
Fr
e
el
ist
l,
Fr
eel
ist
2,
an
d
F
r
eel
ist
3
for
hot
inf
or
m
at
ion
,
wa
rm
inform
at
ion
,
an
d
chill
y
info
rm
a
ti
on
,in
div
i
dual
ly
.
Since
the
ob
st
ru
ct
s
that
store
hot
inf
or
m
at
ion
hav
e
the
m
os
t
el
e
vated
li
kelihood
of
bein
g
era
di
cat
ed,
wh
e
n
ho
t
inf
or
m
at
ion
ar
e
com
po
sed
,
a
fr
ee
s
qu
a
re
is
disp
e
ns
e
d
f
rom
the
pr
i
ncipal
fr
ee
piece
li
st
with
the
m
os
t
m
inim
al
delet
e
ta
l
l
y;
fo
r
war
m
inform
ation
or
ne
w
inf
or
m
at
ion
,
a
fr
ee
sq
ua
re
is
a
pport
ion
ed
from
the
seco
nd
run
down
;
a
nd
for
c
ool
inf
or
m
at
ion
,
a
fr
ee
piece
is
desig
nated
fro
m
the
third
run
dow
n.
On
the
off
ch
ance
that
there
is
no
f
ree
squ
are
in
a
fr
ee
pi
ece
li
st,
a
fr
ee
sq
ua
re
is
dis
pe
ns
e
d
from
the f
ollo
wing
fr
ee
piec
e li
st. Fig
ure
1 shows
the
flo
w
ch
a
rt of
blo
c
k al
locat
ion
.
Fig
ure
1
.
Flo
w
ch
a
rt of Bloc
k al
locat
ion m
eth
od
b)
Create
data bl
ock
li
st
At
the
po
i
nt
wh
e
n
ar
range
m
ents
of
obstr
ucts
that
con
ta
in
inf
or
m
at
ion
are
m
ade
in
RAM,
isolat
e
record
s
are
m
a
defor
hot
in
for
m
at
ion
pieces,
war
m
inf
or
m
a
ti
on
s
qu
a
res,
a
nd
co
ol
in
form
at
ion
pieces.
T
hes
e
three
recor
ds
are
orc
hestrate
d
in
vie
w
of
va
rio
us
crit
eria
for
prof
ic
ie
nt
waste
acc
um
ul
at
ion
a
nd
f
ree
piece
distrib
ution.
D
ue
to
visit
ad
j
ustm
ents,
ho
t
in
form
ation
piec
es
hav
e
a
hi
gh
rate
of
in
valid
pag
es
.
Fr
ee
hin
de
rs
that
ha
ve
the
m
os
t
red
uce
d
del
et
e
ta
ll
y
are
desig
nated
f
or
ho
t
i
nfor
m
at
i
on.
T
hu
sly
,
ho
t
inform
at
ion
sq
ua
res
ought
to
be
vi
ewed
as
first
wh
e
n
waste
ga
therin
g
is
pe
rfor
m
ed.
Ho
t
i
nfor
m
at
ion
squa
r
es
are
rec
orde
d
in
the
dro
pp
i
ng
requ
est
of
t
heir
in
va
li
d
pa
ge
pro
portio
n
to
li
m
it
the
delet
e
f
et
c
hed
am
id
junk
accum
ulati
on
.
Since
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Impr
ovem
e
nt
of
G
arb
age M
anage
men
t f
or
NAN
D
Me
mor
y S
yst
e
m
(
A.
T
hilee
pan
)
541
frosty
inf
or
m
at
ion
a
re d
efi
nitel
y
no
t
re
fr
e
sh
e
d
of
te
ntim
es,
chill
y
inform
ati
on
pieces ha
ve
a
lowe
r
i
nv
al
i
d
pag
e
pro
portion an
d a hi
gh
e
r dele
te
num
ber
. Acco
rd
i
ng
ly
, t
hey ought to
b
e
h
a
ndle
d
la
st am
id trash acc
um
ulati
on
.
2.
GARBA
GE G
ATHERI
NG
METHO
D
YAFFS
pe
rfo
rm
s
trash
acc
um
ulati
on
be
f
or
e
i
nfor
m
at
ion
c
om
po
sing.
Since
t
his
de
ferral
s
the
com
po
se
opera
ti
on
,
re
f
us
e
gat
her
i
ng
ought
to
be
perf
or
m
e
d
w
hile
the
fr
a
m
ewo
r
k
is
out
of
gear
sta
tu
s
to
ke
e
p
the
postp
onem
ent
of
the
co
m
po
se
o
pe
rati
on.
F
or
this
re
aso
n,
tw
o
case
s
are
co
ns
ide
r
ed.
I
n
the
first
place,
base
d
on
t
he
e
vid
e
nce
that
if
the
f
ram
ewo
rk
is
sit
ou
t
of
ge
ar
f
or
2
seco
nd
s,
the
li
kelih
oo
d
that
th
e
f
ram
ewor
k
will
kee
p
on
be
ing
i
ner
t
f
or
4
seco
nds
is
95
%,
the
ju
nk
ga
therin
g
is
pe
rfor
m
ed
w
he
n
th
e
fr
am
ework
is
sit
out
of
gear
for
2
s
econds
.
As
of
now,
t
he
squar
es
with
just
in
valid
pa
ges
a
r
e
erad
ic
at
ed
to
lim
it
the
fr
am
ewor
k
ov
e
r
head.
Sec
ond,
if
t
he
f
ra
m
ewo
r
k
kee
ps
on
run
ning
w
it
ho
ut
sit
out
of
gear
ti
m
e,
ref
us
e
acc
um
ulati
on
is
perfo
rm
ed
wh
e
n
the
qu
a
ntit
y
of
fr
ee
piece
fa
ll
s
ben
eat
h
f
T
this
is
the
qu
a
ntit
y
of
ho
t
in
f
or
m
a
ti
on
pieces
.
Th
e
trash
gathe
rin
g
is
com
plete
d
in
the
re
quest
of
t
he
r
undow
n
w
hose
in
for
m
at
ion
pieces
hav
e
l
ow
e
ra
di
cate
nu
m
ber
s:
hot
inf
or
m
at
ion
li
s
t,
war
m
inform
at
ion
r
undow
n,
an
d
ic
y
infor
m
at
ion
li
st.
To
kee
p
the
e
xtrem
e
delet
e
cost,
the
lim
it
est
ee
m
s
are
set
.
In
t
he
ho
t
in
form
at
io
n
li
st,
the
lim
it
est
ee
m
h
T
is
the
qu
a
ntit
y
of
in
va
li
d
pag
e
s
in
a
piec
e.
In
th
e
eve
nt
that
there
are hot
inf
or
m
at
ion
ob
st
ru
ct
s w
it
h
their
num
ber
of
inv
al
id p
a
ges
m
or
e
no
te
worthy
tha
n
h
T
,
ref
us
e
gathe
rin
g
is
per
f
or
m
ed
in
the
sli
ding
request
of
th
e
inv
al
id
pa
ge
pro
p
ort
ion
in
t
he
ho
t i
nfo
rm
at
io
n
li
st. Fo
r
the
war
m
inf
or
m
ation
li
st, refuse
gathe
rin
g
is pe
rfor
m
ed
for
hi
nd
e
rs
that ha
ve
estee
m
in
of
w
hic
h
is
m
or
e
no
te
w
or
t
hy
than
the
ed
ge
w
T
.
Ultim
a
te
ly
,
in
the
har
s
h
el
e
m
ents
inform
at
ion
li
st,
ru
bb
ish
ac
cum
ulati
on
is
perform
ed
for
the
obstr
ucts
that
hav
e
era
dicat
e
check
s
l
ow
e
r
tha
n
the
lim
it
C
T
.
The
fl
ow
char
t
of
ga
r
bage coll
ect
ion i
s
sh
ow
n
in
F
ig
ure
2.
Fig
ure
2.
Flo
w
ch
a
rt of
garba
ge
c
ollec
ti
on
s
yst
e
m
3.
RESU
LT
S
AND DI
SCUS
S
ION
The pr
opose
d
NAN
D based
garba
ge
c
ollec
ti
on
syst
e
m
is estim
at
ed
and
m
entione
d
i
n
T
a
ble 1.
Table
1.
Propo
sed NA
N
D gar
bag
e
co
ll
ect
io
n sy
stem
Blo
ck
cou
n
t
150
Pag
e cou
n
t per blo
ck
40
File size
0
.5
-
2
0
KB
h
T
20
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vol
.
9
,
No.
3
,
Ma
rc
h
201
8
:
539
–
542
542
The
C
os
t
-
Be
ne
fit ca
lc
ulati
on
has
the l
ongest
w
ast
e ga
theri
ng ti
m
e since it
has
m
or
e subst
antia
l pag
e
s
to
duplica
te
th
an
al
te
rn
at
e
te
chn
i
qu
e
s.
T
he
Gr
ee
dy
cal
cu
la
ti
on
has
le
ss
substanti
al
pa
ges
to
duplica
te
ye
t
inv
est
m
uch
energy
scan
ning
for
the
obj
ec
ti
ve
o
bs
tr
ucts
f
or
the
r
ubbis
h
accum
ulati
on
,
so
it
s
trash
gather
i
ng
tim
e
is
j
us
t
s
om
ewh
at
s
horter
tha
n
t
hat
of
t
he
C
os
t
-
Be
ne
fi
t
cal
culat
ion
.
T
he
c
om
par
ison
diag
ram
of
propose
d
and co
nventi
onal
N
A
N
D gar
ba
ge base
d
m
e
m
or
y dev
ic
e is
sh
ow
n
in
F
ig
ure
3.
Fig
ure
3
.
Com
par
is
on of
garb
age
value
4.
CONCL
US
I
O
N
Existi
ng
t
rash
gathe
rin
g
te
ch
niques
need
t
o
look
f
or
hinde
r
s
that
ha
ve
the
ideal
delet
e
ta
ke
n
a
toll
in
the
wa
ke
of
de
ci
din
g
the
er
adicat
e
cost
of
each
piece
be
fore
erasi
ng
sq
ua
res
.
I
n
thi
s
m
ann
er,
the
refuse
accum
ulati
on
tim
e
incre
m
ents
in
extent
to
t
he
blaze
m
e
mo
ry
est
im
a
te
.
Be
sides,
they
don'
t
dem
on
strat
e
gr
eat
execu
ti
on
f
or
wear
le
velin
g
i
n
li
ght
of
t
he
f
act
that
the
pr
e
sence
of
hot
further
m
or
e,
ic
y
inf
or
m
at
ion
ca
n'
t
be
ref
le
ct
ed
.
At
la
st,
a
powe
rful
fr
ee
piece
desi
gn
at
io
n
strat
eg
y
in
li
ght
of
i
nfo
rm
ation
s
ort
is
pro
posed
f
or
wea
r
le
veling
of
al
l
squares
.
Try
com
es
abo
ut
dem
on
strat
e
tha
t
the
sta
ndar
d
dev
ia
ti
on
of
t
he
delet
e
ta
ll
y
i
s
low
payi
ng
li
tt
le
m
ind
to
the
gli
m
m
er
m
e
m
or
y
us
e.
In
e
xpansi
on,
the
re
f
us
e
a
ccum
ulati
on
tim
e
is
abb
rev
ia
te
d
by
decr
easi
ng the
ideal
oppo
rtu
nity
f
or hu
nting
dow
n
the
ob
j
ec
ti
ve
pieces
f
or
the w
a
ste
g
at
he
rin
g.
REFERE
NCE
S
[1]
Chang,
Li
-
Pin,
a
nd
Te
i
-
W
e
i
Kuo
.
An
adaptive
str
ipi
ng
archi
tecture
for
f
lash
memor
y
storage
syste
ms
of
embe
dde
d
systems.
In
Re
al
-
Ti
m
e
and
Embe
dded
T
ec
hnolog
y
and
Appli
catio
ns
S
y
m
posium
,
2002.
Proce
edi
n
gs.
Ei
gh
th
IE
EE,
pp.
187
-
196
.
I
E
EE
,
2002.
[2]
Le
e
,
Sungjin,
D
ongkun
Shin,
Y
oung
-
Jin
Kim
,
and
Jihong
Kim
.
LAST
:
Locali
ty
-
aware
sec
tor
translati
ons
for
NAND
fl
ash me
mor
y
-
bas
ed
storage
systems.
ACM
SIG
OPS Ope
ra
ti
ng
S
y
st
ems
Revi
ew
42
,
no
.
6
(2008):
36
-
42.
[3]
Chang,
Li
-
Pin,
Te
i
-
W
ei
Kuo,
a
nd
Shi
-
W
u
Lo.
Re
al
-
ti
me
garba
ge
co
ll
e
ct
ion
for
fl
ash
-
memor
y
s
torage
systems
of
real
-
ti
me
embe
d
ded
systems.
AC
M
Tra
nsac
ti
ons
on
Embedde
d
C
om
puti
ng
S
y
ste
m
s
(TE
CS
)
3,
n
o.
4
(2004)
:
837
-
863.
[4]
Chang,
Yuan
-
Hao,
Jen
-
W
ei
Hs
ie
h,
and
Tei
-
W
ei
Kuo.
Endurance
enhanc
eme
n
t
of
fl
ash
-
memor
y
storage
systems
:
an
ef
fici
ent
stati
c
wear
leve
l
ing
design.
In
Proce
edi
ngs
of
the
44t
h
annua
l
Design
Autom
at
ion
Confer
ence,
pp
.
2
12
-
217.
ACM
,
2007
.
[5]
Horn,
Robert
L
.
Priorit
y
-
based
g
arbage
col
l
ec
t
io
n
for
data
storage
systems.
U.S.
Pate
nt
9
,
430,
37
6,
issued
Augus
t
30,
2016
.
[6]
Guan,
Yong,
Guohui
W
ang,
Chenl
in
M
a,
R
en
hai
Ch
en,
Yi
W
ang,
and
Zili
Shao.
A
B
lock
-
Leve
l
Log
-
Bl
oc
k
Manage
ment
S
c
hem
e
for
MLC N
AND
Fl
ash M
em
ory
Storage
S
yst
ems.
IE
EE
Tra
ns
ac
t
ions o
n
Com
pute
rs (2017)
.
[7]
Ro
y
,
S.K.
,
Shar
an,
P.,
Nal
ini
,
R.
and
Sriniva
s,
T.
,
2015,
Nove
m
ber
.
An
ef
fici
e
nt
design
of
serial
and
paralle
l
memor
y
using
Quantum
dot
cellul
ar
automata.
In
TE
NCO
N
2015
-
20
15
IEE
E
Regi
on
10
Confer
ence
(pp
.
1
-
4
).
IEE
E
.
[8]
Sundara
Raman
,
R.
,
Sankar
a
Na
ra
y
ana
n
&
G.,
Manoha
ra
n,
N.
,
Smar
t
memor
y
all
o
ys
as
struc
tural
composit
e
s
,
Inte
rnational
Jo
urnal
of Appl
i
ed Engi
ne
ering
R
ese
arch
,
vo
l. 9, no. 18, pp. 3939
-
39
48,
2014
.
[9]
Sa
y
y
ad
,
R.
and
Redka
r,
S.,
20
16.
Fail
ure
Ana
l
y
sis
and
Rel
i
a
bil
ity
Stud
y
of
NA
ND
Flash
-
B
ase
d
Solid
Stat
e
Drive
s.
Indon
esi
an
Journal
o
f El
ec
tri
cal
Engi
n
eer
ing
and
Compu
te
r Science
,
2
(2)
,
pp
.
315
-
327.
[10]
W
ei
,
S.
,
Da
i,
L.
and
Zh
ang,
J.,
2
014.
Flex
ibl
e
Na
nofa
bricat
ion
Eq
uipment:
E
-
beam
Li
thogra
ph
y
S
y
stem
Based
o
n
SEM.
Indone
sia
n
Journal
o
f El
e
ct
rical E
ng
ineering
and
Compute
r Sc
ie
n
ce
,
12
(5)
,
pp.
3841
-
3848
.
[11]
Babuprasanth.
V,
“
Cloud
Connec
te
d
Sm
art
Gas
Le
ak
age
Det
ec
t
i
on
and
Safe
t
y
P
re
ca
u
ti
on
S
y
st
e
m
”,
Inte
rnation
al
Journal
of
MC
s
quare
Scientific
Re
search
(
IJ
MS
R)
.
2014;6(1):
66
-
74.
Evaluation Warning : The document was created with Spire.PDF for Python.